The Frontier AI Safety Act could ensure AI companies are committing more resources to safeguard their systems. The legislation would require major AI developers to dedicate a percentage of their research to safety, publicly report those investments, and undergo independent verification.
The bill was introduced as Anthropic, is set to have the largest stock market debut ever after filing for an initial public offering this week. Democratic State Senator Josh Becker, who authored the bill, said AI poses an "existential risk" with those who interact with it, something the company itself acknowledges.
"We don't always know why it does what it does," said Becker. "If there was a 10% chance that I board an airplane that we didn't know if it was going to crash, would people want to board that that airplane?"
Becker is hosting the second part of a two-day forum with AI experts this week intended to explore what safety research is being done, what gaps remain, and other accountability measures that need to be taken.